Hello, finally I solved all issues and received the following performance values between a CIFS share on OpenSolaris and Windows Vista: Write Vista to OpenSolaris: 35,6 MByte/s Read Vistra from OpenSolaris: 34,5 MByte/s
I made the measurement with a 2048 MByte and the H2testw tool. (http://www.heise.de/software/download/h2testw/50539) Used HW: - Intel Motherboard D945GSEJT - Intel Atom CPU N270 1.6Ghz (32Bit) - Intel Mobile 945GSE Express Chipset - Intel 82945GSE Graphics and Memory Controller Hub (GMCH) - Intel 82801GBM I/O Controller Hub (ICH7-M) - Realtek 8111DL Gigabit Ethernet Controller - 2 Gb Ram DDR2 533 Mhz - 2 x 1TB Western Digital Caviar Green (WD10EADS) Used SW: OpenSolaris 2009/06 Configuration as described in http://www.h-online.com/open/OpenSolaris-as-a-file-server--/features/112212 Problems: 1. Performance of OpenSolaris rge network driver. 2. Handling of incomming data within Windows Vista. Solution. 1. Instalation of gani driver for Realtek rtl8169 rtl8110 rtl8168 rtl8101 PCI/PCI-Express GbE chipset. http://homepage2.nifty.com/mrym3/taiyodo/eng/ The driver works as well for the rtl8111cl mounted on the board. Note: First remove all rge entries in gedit /etc/driver_aliases and reboot before try to install. Now everithin on OpenSolaris performs as it should and you have good transfere rates e.g. towards Windows XP (No changes on powersafe necessary!) 2. Fix Vista! With Vista Microsoft has introduced some automatic tuning posibilities within TCP stack. This should be already solved within some service packs but where still present on my Vista with SP2. For me it helped to disable everythin possible and reboot both Vista and Solaris. Execute from cmd windows with admin rights netsh interface tcp set global rss=disabled chimney=disabled autotuninglevel=disabled congestionprovider=none ecncapability=disabled timestamps=disabled Thanks to all and especially to jkeil. Best regards, Oliver -- This message posted from opensolaris.org _______________________________________________ opensolaris-discuss mailing list [email protected]
